JPH0623983B2 - Distortion correction device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Outline] Linear distortion of a graphic image including a grid obtained from a graphic drawn with a grid as a reference is corrected based on a reference point set at a predetermined position in the graphic, The vicinity of the grid point position of the corrected graphic image is extracted, and the non-linear distortion of the image of only the graphic obtained from the graphic is corrected for the extracted region. Thus
The reduction in recognition rate caused by the non-linear expansion and contraction of the paper on which the graphic is drawn can be eliminated.

[Industrial application field]

The present invention relates to a distortion correction device, and more particularly to a distortion correction device capable of eliminating a problem caused by a non-linear expansion / contraction of a paper on which a graphic is drawn.

There is a system (CAD system) that processes an image obtained by automatically reading a drawing target having elasticity such as a design drawing drawn on paper. In such a system, no problems occur as long as the image input to the system faithfully represents the original figure.

However, in the above-mentioned paper and the like, in addition to the linear distortion and the rotational distortion due to the image input device, nonlinear distortion may be given to the image input to the system due to its elasticity and the like. Since these distortions are one of the factors that lower the image recognition rate, it becomes necessary to input an image from which such distortions have been removed to the image recognition processing unit.

[Conventional technology]

Conventionally, a design drawing drawn when a printed board pattern diagram or the like is input to a CAD system is handwritten on the basis of a grid on a sheet set corresponding to the coordinate system of the CAD system.

As a method of inputting the design drawing drawn on the paper in this way into the CAD system, the present applicants have filed Japanese Patent Application No.
-96758 (graphic processing method) proposed a method that can automatically recognize a design drawing on a paper by using a small rectangular area based on a grid as a processing unit.

However, if you try to process drawings with this method,
The lattice spacing of the image data obtained from the image pickup means is influenced by the expansion and contraction of the sheet, and therefore, "resolution of the image pickup means (for example, 8 lines / mm)" × "interlattice distance on the sheet (for example, 2 mm)" The number of pixels is not always ", that is, distortion may occur.

The present applicant has already proposed a technique for correcting such distortion in Japanese Patent Application No. 54-97613. FIG. 6 is an explanatory view for explaining the technique, and FIG. 6 (a)
Is the original size of the input drawing, and FIG. 6 (b) is the drawing in the memory space when distortion occurs. The following approximate expression holds between them.

However, ΔX = X ₃ −X ₁ and ΔY = Y ₃ −Y ₄ . According to the above equation, the address (X ′, Y ′) in the memory space is the dimensions X _W , Y _W of the original drawing, the address (X, Y) in the memory space when no distortion occurs, and the case where further distortion occurs. Can be corrected using the addresses (X ₁ , Y ₁ ) to (X ₄ , Y ₄ ) on the drawing. This makes it possible to correct the rotational distortion when a drawing image is input from an observation device such as a facsimile or a drum scanner. Further, the distortion due to expansion and contraction of the drawing can be corrected as long as the distortion is linear if the actual number of grids on the drawing is known. Therefore, in this technique, reference marks are drawn in advance at the four corners of the drawing, the marks are detected by a monitor system, and they are used for distortion correction.

[Problems to be solved by the invention]

Although it is possible to correct the input drawing image by the latter method described above, distortion that cannot be corrected by that method may be included in the image to be processed. It arises from the tendency of the paper to undergo non-linear expansion and contraction as the size of the design increases. Such distortion is a serious problem in the drawing process using the processing area with the grid points as a reference, and is one of the major factors that reduce the recognition rate.

The present invention has been made in view of the above technical problems, and provides a distortion correction device that can correct not only linear distortion but also nonlinear distortion and that contributes to an improvement in recognition rate in graphic processing. Is its purpose.

[Means for solving problems]

FIG. 1 shows the principle of the present invention. In the figure, reference numeral 2 denotes an image input means, which obtains an image of only the figure and a figure image including the lattice from the figure drawn from the figure 1 with reference to the lattice of the object to be drawn 1 according to the present invention. It is for inputting into the processing system. Reference numeral 3 denotes a distortion correction unit that corrects the rotational distortion of a graphic including a lattice and the linear distortion due to the expansion and contraction of the drawing object based on a reference point set at a predetermined position in the drawing object 1.
Reference numeral 4 denotes a non-linear distortion correction means, which corrects the non-linear distortion of the image of only the figure near the lattice point position corrected by the distortion correction means 3.

[Action]

From the figure drawn with the lattice of the drawing object as a reference, an image of only the figure and a figure image including the lattice are input by the image input means 2.
Obtained through. The rotation distortion of the figure including the lattice thus obtained and the linear distortion due to the expansion and contraction of the object to be drawn are corrected by the distortion correcting means 3 based on the reference point set at a predetermined position in the object to be drawn. It Then, the non-linear distortion is corrected by the non-linear distortion correcting means 4 for the image of only the figure near the lattice point position corrected by the distortion correcting means 3, and the figure image from which the distortion is removed is subjected to the recognition processing. It

In this way, it is possible to correct not only the rotational distortion that occurs when the image is read by the image input device and the linear distortion due to the expansion and contraction of the object to be drawn, but also the nonlinear distortion, so that the processing accuracy or recognition rate in the graphic processing is improved. Can be achieved.

〔Example〕

FIG. 2 shows an embodiment of the present invention. In this figure, 1
Reference numeral 0 is a design drawing sheet, on which grids are drawn at predetermined intervals, and the design drawing is drawn on the basis of the grids. The color for drawing the design and the color for the grid are different. The design drawings drawn in different colors are read by the observation device 11 having a color separation function. A graphic image including two types of read images, that is, a graphic (object to be processed) 12 and a reference point 13 (see FIG. 3).
(b)) and the graphic image including the graphic 12, the grid 14 and the reference point 13 (FIG. 3 (a)) are stored in the corresponding image memories ((I), (II)) 14 and 15, respectively. To be done.

Next, in the reference point detection circuit 16, the technique proposed by the inventors of the present applicant described above (Japanese Patent Application No. 54-9).
6758), the reference point address of the graphic image stored in the image memory (II) 15 is detected, and the table for distortion correction is stored in the address control unit 17. The address control unit 17 sequentially reads the image data near the grid points based on the table. The grid coordinates of the read data are provisionally determined under the above-described technique, that is, the linear distortion due to the rotational distortion and the expansion / contraction of the paper is corrected.

The data in which the linear distortion is corrected, that is, the image data in the vicinity of the lattice points (see FIG. 4 (a)) is supplied to the minute distortion correction circuit 18. The following processing is performed in the micro-distortion correction circuit 18. For input image data in a rectangular area as shown in FIG. 4 (a), 1 × as shown in FIG. 4 (b)
The n, n × 1 primary windows 19 and 20 are run vertically and horizontally from the center of the rectangular area (coordinates of grid points tentatively determined by the technique disclosed in Japanese Patent Application No. 54-96758) and enter the window. A position where the number of black pixels is equal to or larger than a predetermined threshold is extracted. Thereby, for example, the minute distortion amounts ΔX and ΔY as shown in FIG. 4B are detected for the image of the rectangular area as shown in FIG.

These ΔX and ΔY are passed to the address control unit 17, and the graphic image of the image memory 14 is read so as to correct the grid point coordinates provisionally determined as described above by the minute distortion amounts ΔX and ΔY. In this way, the non-linear distortion is corrected and the accurate image data of the design drawing (object to be processed) can be cut out and sent to the graphic processing unit (not shown). An example of the cut out is shown in FIG.

The object to be processed may be a pattern other than the design drawing, for example, a character.

〔The invention's effect〕

As described above, according to the present invention, the rotational distortion that occurs when an image is read by the image input device, the linear distortion that is caused by the expansion and contraction of the object to be drawn, and the nonlinear distortion are also corrected. The accuracy or recognition rate can be improved.

[Brief description of drawings]

 FIG. 1 is a principle diagram of the present invention, FIG. 2 is a diagram showing an embodiment of the present invention, FIG. 3 is an explanatory diagram of color separation used in the present invention, and FIG. 4 is an explanatory diagram of minute distortion extraction processing, FIG. 5 is a diagram showing an example of minute distortion correction, and FIG. 6 is a diagram for explaining a conventional distortion correction method. In FIG. 1, 1 is an object to be drawn, 2 is image input means, 3 is distortion correction means, and 4 is non-linear distortion correction means.

Claims (2)

[Claims] 1. A processing apparatus for an object to be processed drawn with reference to the grid of an object to be drawn (1), wherein a first image of only the object to be processed and a second image containing the object and the grid are processed. Image input means that can input the image of (2)
And a distortion correction means (3) for correcting the rotational distortion / linear distortion of the second image based on a reference point set at a predetermined position of the object to be drawn.
And a non-linear distortion correcting means (4) for correcting the non-linear distortion of the first image near the position of the grid point corrected by the distortion correcting means, . 2. The distortion correction apparatus according to claim 1, wherein the object to be processed is a design drawing, and the design drawing is drawn with a grid of a sheet as a reference.